Visualization and Analysis of Urban Taxi Demand Survey Data

The focus of this Yokohama National University research project was to develop an efficient urban taxi management system using 417 taxis with car-mounted Global Positioning System (GPS) units in Sendai City, Japan. Demand estimates of taxis are dependent on time and spatial factors. Taxi demand is expressed in terms of customer population, average travel time, and purpose of travel. This is compared to regional data such as distances from train stations, building densities, and population densities. Demand is estimated for different times of the day and days of the week using regression analysis. To visualize and analyze these factors, geoprocessing tools such as kernel density, zonal statistics, and principal components were used.

The balance of supply and demand was assessed using an evaluation index. This index expresses how much time is required to reach each zone. ArcGIS Network Analyst calculates the destination of taxis in a current zone and determines an optimum route that will maximize the probability that a customer will be waiting in that zone. Not only does this create a more efficient form of dispatching but also minimizes the amount of time wasted searching for customers and ultimately cuts down exhaust emissions.

Courtesy of Kentaro Taguchi, Satoshi Yoshida, and Satoru Sadohara, Yokohama National University.